Carbon isotope fractionation of amino acids in fishmuscle reflects biosynthesis and isotopic routingfromdietary protein

KeltonW.McMahon1*, Marilyn L. Fogel2, Travis S. Elsdon3 and SimonR. Thorrold4

1Massachusetts Institute of Technology andWoodsHole Oceanographic Institution, Joint Program inOceanography and

Ocean Engineering,WoodsHole Oceanographic Institution,WoodsHole, MA 02543, USA; 2Carnegie Institution of

Washington, 5251 Broad BranchRd. N.W.,Washington, DC 20015, USA; 3Southern Seas Ecology Laboratories, School of

Earth and Environmental Sciences, University of Adelaide, Adelaide, SA 5005, Australia; and 4Biology Department,Woods

Hole Oceanographic Institution,WoodsHole, MA 02543, USA

Summary

1. Analysis of stable carbon isotopes is a valuable tool for studies of diet, habitat use and

migration. However, significant variability in the degree of trophic fractionation (D13CC-D)between consumer (C) and diet (D) has highlighted our lack of understanding of the biochemicaland physiological underpinnings of stable isotope ratios in tissues.

2. An opportunity now exists to increase the specificity of dietary studies by analyzing the d13Cvalues of amino acids (AAs). Common mummichogs (Fundulus heteroclitus, Linnaeus 1766) were

reared on four isotopically distinct diets to examine individual AA D13CC-D variability in fishmuscle.

3. Modest bulk tissue D13CC-D values reflected relatively large trophic fractionation for manynon-essential AAs and little to no fractionation for all essential AAs.

4. Essential AA d13C values were not significantly different between diet and consumer(D13CC-D = 0Æ0 ± 0Æ4&), making them ideal tracers of carbon sources at the base of the food

web. Stable isotope analysis of muscle essential AAs provides a promising tool for dietaryreconstruction and identifying baseline d13C values to track animal movement through isotopi-cally distinct food webs.

5. Non-essential AA D13CC-D values showed evidence of both de novo biosynthesis and directisotopic routing from dietary protein. We attributed patterns in D13CC-D to variability in protein

content and AA composition of the diet as well as differential utilization of dietary constituentscontributing to the bulk carbon pool. This variability illustrates the complicated nature of metabo-

lism and suggests caution must be taken with the assumptions used to interpret bulk stable isotopedata in dietary studies.

6. Our study is the first to investigate the expression of AA D13CC-D values for a marine vertebrateand should provide for significant refinements in studies of diet, habitat use and migration usingstable isotopes.

Key-words: compound-specific stable isotope analysis, feeding experiment, food web, metabolic

processing, trophic dynamics

Introduction

Stable isotope analysis (SIA) has become a routine tool in

ecology for studies of diet and trophic dynamics (Peterson &

Fry 1987; Gannes, Martınez del Rio & Koch 1998; Michener

& Kaufmann 2007), habitat use (McMahon, Johnson &

Ambrose 2005; Cherel et al. 2007) and animal migration

(Hansson et al. 1997; Hobson 1999; Rubenstein & Hobson

2004). Bulk tissue SIA studies using carbon rely upon the

assumption that the isotope composition of a consumer

reflects the weighted average of the carbon isotope composi-

tions of its diet with a small amount of diet (D) to consumer

(C) fractionation, hereafter D13CC-D (typically 0–1&,

DeNiro & Epstein 1978; Fry, Joern & Parker 1978). Despite*Correspondence author. E-mail: kmcmahon@whoi.edu

Journal of Animal Ecology 2010, 79, 1132–1141 doi: 10.1111/j.1365-2656.2010.01722.x

! 2010TheAuthors. Journal compilation! 2010 British Ecological Society

the prevalence of bulk SIA in ecological studies of diet and

food webs, there are still a number of confounding factors

that can complicate interpretations of bulk SIA data.

The carbon isotope composition at the base of the food

web (d13Cbase) ultimately determines the d13C values of

higher trophic level consumers. Without suitable estimates of

d13Cbase, which can vary both spatially and temporally (Van-

der Zanden &Rasmussen 1999; Graham et al. 2009), it is dif-

ficult to interpret consumer d13C values using bulk SIA in

light of potential changes in food web structure vs. variations

in d13Cbase (Post 2002). This can be particularly problematic

when studying the diet and trophic dynamics of highly migra-

tory marine organisms that move among isotopically distinct

food webs (Estrada, Lutcavage & Thorrold 2005; Graham

et al. 2009). There can also be significant variation in D13CC-

D, ranging from )3 to +5&, depending on consumer taxa,

diet and tissues analysed (Gannes, O’Brien & Martınez del

Rio 1997; Vander Zanden & Rasmussen 2001; McCutchan

et al. 2003; Olive et al. 2003). Furthermore, the d13C value of

consumer tissue may not always follow bulk diet d13C values

because the carbon skeletons of different dietary constituents

(proteins, lipids and carbohydrates) can be routed to differ-

ent tissue constituents (‘isotopic routing’; Schwarcz 1991).

Several studies have emphasized the problems that isotopic

routing poses to the interpretation of stable isotope data in

diet reconstructions (Parkington 1991; Schwarcz & Schoen-

inger 1991; Ambrose & Norr 1993). All of these factors can

make interpretations of bulk tissue SIA challenging for stud-

ies of diet and migration, prompting a call for studies that

examine the biochemical and physiological basis of stable

isotope ratios in ecology (Gannes, O’Brien & Martınez del

Rio 1997; Gannes, Martınez del Rio & Koch 1998; Karasov

&Martınez del Rio 2007).

The opportunity now exists to increase the specificity of

dietary studies by analysing d13C values of specific biochemi-

cal compounds, including amino acids (AAs), thanks to

recent advances in gas chromatography-combustion-isotope

ratio monitoring mass spectrometry (GC-C-irmMS)

(Merritt, Brand & Hayes 1994; Meier-Augenstein 1999;

Sessions 2006). Stable isotope analysis of individual AAs has

the potential to provide more detailed information about diet

(Fantle et al. 1999; Fogel & Tuross 2003; Popp et al. 2007)

and the sources of complex mixtures of organic matter (Uhle

et al. 1997; McCarthy et al. 2004) than conventional bulk

tissue SIA.

There have been very few controlled feeding experiments

examining the trophic fractionation of individual AAs

between diet and consumer (Hare et al. 1991; Howland et al.

2003; Jim et al. 2006). Studies to date found modest bulk tis-

sue D13CC-D values (!1&) actually reflected an average of

relatively large fractionations in many non-essential AAs and

comparatively little fractionation in most essential AAs.

However, there was considerable variation in D13CC-D across

diets and individual AAs among studies. Furthermore, these

studies all dealt with terrestrial vertebrates (pigs and rats),

yet no controlled feeding experiments looking at compound-

specific D13CC-D have been conducted on an aquatic

vertebrate. Given the variability in bulk tissueD13CC-D across

terrestrial and aquatic taxa (Vander Zanden & Rasmussen

2001; McCutchan et al. 2003), it is important to determine

the mechanisms leading to variability in the fractionation of

AA d13C values for aquatic taxa.

We reared common mummichogs (Fundulus heteroclitus,

Linnaeus 1766) on four isotopically distinct diets to examine

trophic fractionation (D13CC-D) of individual AAs between

diet and consumer. By choosing an herbivorous diet, two car-

nivorous diets and an omnivorous diet, we aimed to examine

the potential variability in AA D13CC-D. We addressed the

specific question: What is the isotopic relationship between

diet and consumer for individual AAs in fish muscle? We

focused on muscle tissue, because it is one of the most com-

monly used tissues in ecological studies of diet and trophic

dynamics. We hypothesized that non-essential AA d13C val-

ues would show evidence of both de novo biosynthesis and

direct isotopic routing from dietary protein while essential

AA d13C values would only reflect isotopic routing. Similar

results were found for pigs (Hare et al. 1991; Howland et al.

2003) and rats (Jim et al. 2006), although the magnitude and

direction of trophic fractionation was quite variable. We also

hypothesized that fish fed a high protein content diet would

exhibit a greater degree of isotopic routing because routing is

thought to be more efficient than de novo biosynthesis when

non-essential AAs are sufficiently available (Ambrose &

Norr 1993; Tieszen & Fagre 1993; Jim et al. 2006). Finally,

we predicted that a deficit in non-essential AA abundance in

diet relative to consumer tissue would result in higher trophic

fractionation than would be expected from diets with excess

non-essential AAs due to enhanced biosynthesis. Our study

is the first to investigate the expression of individual AA

D13CC-D values for an aquatic, vertebrate and should provide

significant refinements to studies of diet, habitat use and

migration using stable isotopes.

Materials andmethods

FEEDING EXPERIMENT

We conducted a controlled feeding experiment on juvenile mummic-

hogs (Fundulus heteroclitus) reared at the Atlantic Ecology Division,

U.S. Environmental Protection Agency, in Narragansett, Rhode

Island, USA. Adult F. heteroclitus collected from a salt marsh in

Sandwich, Massachusetts were held in flow through seawater at tem-

peratures elevated above ambient to induce spawning. Eggs from

two spawnings were collected and transferred to tanks and allowed

to hatch. Juvenile fish were reared on an Artemia diet for 6 weeks

(approximate length: 11 mm), after which they were transferred to

experimental tanks.

Experimental tanks consisted of twelve 40 gallon aquaria with flow

through seawater at ambient temperature (20 "C) in two randomly

positioned rows under a 12 : 12 h light : dark cycle from fluorescent

tubes. Twenty juvenile F. heteroclitus were placed in each tank. Die-

tary manipulations consisted of triplicate tanks of fish reared on one

of four isotopically distinct diets. Aplant-based commercial fish pellet

(Vegi-Pro, FreedonFeeds Inc., Urbana, OH,USA) consisted primar-

ily of wheat and soy with a small contribution from corn meal. A

Carbon isotope fractionation of fish muscle amino acids 1133

! 2010 TheAuthors. Journal compilation! 2010 British Ecological Society, Journal of Animal Ecology, 79, 1132–1141

second commercial fish pellet (Bio-Vita, Bio-Oregon, Westbrook,

ME,USA)consistedoffishandkrillmeal,wheatglutenandwheypro-

tein. Two natural animal-based diets, squid and clam, were obtained

from a local supermarket, homogenized and then freeze-dried before

being fed to the experimental fish. Proximate analysis of moisture by

loss on drying at 135 "C for 2 h (method 930Æ15; AOAC 2005), crude

protein by combustion (method 990Æ03; AOAC 2005), crude fat by

ether extraction (method 920Æ39; AOAC 2005), crude fibre (method

978Æ10; AOAC 2005) and ash (method 942Æ05; AOAC 2005) were

conducted on all four diets at the New Jersey Feed Laboratory,

Trenton, New Jersey, USA. Carbohydrate content was determined

as 100%minus the sum ofmoisture, protein, fat and ash. Amino acid

compositions (16 individual AAs) (method 994Æ12; AOAC 2005)

of the four diets and fish muscle from each treatment were also

determined at theNewJerseyFeedLaboratory (n = 3replicates).

SAMPLE PREPARATION AND ANALYSIS

Fish were fed to saturation once per day, and tanks were cleaned of

excess food and bio-films every 3 days. Fish were reared on isotopi-

cally distinct diets for 8 weeks and more than doubled in biomass

during that time. All fish were then sacrificed in an ice slurry, frozen

and freeze dried for 72 h. White muscle was removed from each fish,

homogenized using a mortar and pestle, and subdivided into two

portions, one for bulk tissue SIA and the other for compound-

specific SIA. Results from the bulk tissue analyses are presented

elsewhere (Elsdon et al. 2010).

Approximately 2 mg of each sample, both diet and fish muscle,

were acid hydrolysed in 1 mL of 6 nHCl at 110 "C for 20 h to isolate

the total free AAs. Samples were neutralized with ultrapure water

and evaporated to dryness via rotary evaporation to remove HCl

before being resuspended in 1 mL of ultrapure water. Samples were

then passed through solid phase extraction-C18 columns to remove

particulates and melanoidins. After drying under a stream of N2 gas,

the total free AAs were derivatized by esterification with acidified iso-

propanol followed by acetylation with trifluoroacetic anhydride

(Silfer et al. 1991). The resulting derivatized AAs were diluted to a

concentration of 2 lg lL)1 in dichloromethane.

Approximately 2 lg of AAs (via 1 lL injection) were injected on

column in splitless mode at 220 "C and separated on an HP Ultra-1

column (50 m length, 0Æ32 mm inner diameter and 0Æ52 lm film

thickness; Hewlett Packard, Wilmington, DE, USA) in a Varian

3400 Gas Chromatograph (GC) at the Carnegie Geophysical Labo-

ratory, Washington, DC, USA. Gas chromatography conditions

were set to optimize peak separation and shape as follows: initial tem-

perature 75 "C held for 2 min; ramped to 90 "C at 4 "C min)1, held

for 4 min; ramped to 185 "C at 4 "C min)1, held for 5 min; ramped

to 250 "C at 10 "C min)1, held 2 min; ramped to 300 "C at

20 "C min)1, held for 8 min. The separated AA peaks were combu-

sted in a FinneganGC continuous flow interface at 980 "C, thenmea-

sured as CO2 on a Finnegan MAT DeltaPlusXL or Delta V

Advantage isotope ratiomass spectrometer. Twelve of the 16 individ-

ual AAs identified had sufficient baseline separation for stable carbon

isotope analysis, accounting for c. 80% of the total AA per cent

abundance. Glutamic acid and aspartic acid peaks contained

unknown contributions from glutamine and asparagine, respectively,

due to conversion to their dicarboxylic acids during acid hydrolysis.

For consumer muscle, three replicate tanks were analysed per treat-

ment, with three fish analysed per tank. Three replicate samples of

each of the four diets were analysed following the same procedure as

the fish muscle. All compound-specific SIA samples were analysed in

duplicate along withAA standards of known isotopic composition.

DATA ANALYSIS

Stable isotopes are expressed in standard delta (d) notation accordingto the equation

d13CSample " #$13C=12Csample

13C=12Cstandard% & 1' ( 1000:

Trophic fractionation factors (D13CC-D) were calculated for each

treatment as D13CC-D = d13CC)d13CD, where d13CC and d13CD rep-

resent the d13C values of the consumer and diet respectively. Stan-

dardization of runs was achieved using intermittent pulses of a CO2

reference gas of known isotopic value.

To correct for the introduction of exogenous carbon and kinetic

fractionation during derivatization (Silfer et al. 1991), AA standards

of known isotopic value were derivatized and analysed concurrently

with the samples. Error for determining the isotopic composition of

the exogenous carbon added during derivatization averaged±0Æ4&.

Differences in bulk d13C of diet and fish muscle among treatments

were assessed using separate one-way analyses of variance (anovas)

and Tukey’s HSD post-hoc tests (a = 0Æ05). Differences in individual

AA d13C values within and among treatments for both diet and fish

muscle were determined using separate model I (treatment and AA

factors fixed) two-way anovas and Tukey’s HSD post-hoc tests

(a = 0Æ05). To examine differences in individual AA D13CC-D both

within and among treatments, AAs were a priori subdivided into

non-essential and essential AAs. Differences in non-essential and

essential AA D13CC-D values were analysed using separate model I

(treatment and AA factors fixed) two-way anovas and Tukey’s

HSD post-hoc tests. Separate two-sided one sample t-tests were

used to determine if AA D13C values were significantly different

from 0&. Linear regressions were performed to compare AA d13Cvalues in muscle (d13Cmuscle_AA) to (i) their respective dietary AAs

(d13Cdiet_AA) and (ii) the bulk diets (d13Cbulk_diet). Using the AA com-

position data, we determined the difference in AA per cent abun-

dance between diet and muscle, with negative values indicating a

deficit in AA abundance in diet relative to muscle. To look for trends

in trophic fractionation as a function of AA composition, we

conducted a correlation analysis between the AA per cent abundance

difference and AA D13CC-D for all AAs showing de novo biosynthesis

(D13CC-D significantly different from 0&). All statistics were per-

formed in prism version 4.0.

Results

Bulk d13C values were significantly different among the diets

(one-way anova, F = 717Æ7, P < 0Æ05; Fig. 1) and fish mus-

cle (one-way anova, F = 321Æ8, P < 0Æ05; Fig. 1) across

treatments. The Vegi-Pro diet had the highest carbohydrate

content (73%) and lowest crude fat (6%) and protein (8%)

content, while the squid and clam diets had the highest crude

fat (18%) and protein (69% and 71% respectively) content

and almost no carbohydrates (Table 1). Bio-Vita content

was generally intermediate between the Vegi-Pro and the ani-

mal-based diets, with the exception of a high crude fat con-

tent (24%).

The mean range in d13C values across all 12 AAs analysed

was 27Æ9 ± 6Æ9& for diet (Fig. 1a) and 23Æ6 ± 2Æ6& for fish

muscle (Fig. 1b). We found significant differences in dietary

d13C values (Fig. 1a) among individual AAs (two-way anova,

d.f. = 11, 96, F = 1239Æ0, P < 0Æ05) and among diet treat-

1134 K.W.McMahon et al.

! 2010 TheAuthors. Journal compilation! 2010 British Ecological Society, Journal of Animal Ecology, 79, 1132–1141

ments (d.f. = 3, 96, F = 552Æ0, P < 0Æ05). However, vari-

ability in AA d13C values was not consistent among diet

treatments, generating a significant diet · AA interaction

(d.f. = 33, 96, F = 21Æ71, P < 0Æ05). Fish muscle AA d13Cvalues (Fig. 1b) showed similar patterns to those of the diets,

with significant differences among individual AAs (two-way

anova, d.f. = 11, 96, F = 2681Æ0, P < 0Æ05) and among diet

treatments (d.f. = 3, 96, F = 642Æ7, P < 0Æ05), including a

significant interaction term (d.f. = 33, 96, F = 18Æ72,P < 0Æ05).Despite significant variability in individual AA d13C values

in diet and muscle, there were several consistent patterns in

our data. All AAs from the Vegi-Pro treatment, both diet

and fish muscle, were the most 13C-depleted, while AAs from

the squid and clam treatments were typically the most 13C-

enriched. Glycine and serine were always the most 13C-

enriched AAs in all treatments for both diet and fish muscle,

where as valine, phenylalanine and leucine were always the

most 13C-depleted AAs in all treatments. The d13C value

of aspartic acid, glutamic acid and proline were generally

similar to one another in each treatment for both diet and

muscle, although the values diverged more so for the Vegi-

Pro diet. Finally, the non-essential AAs were 13C-enriched

relative to the essential AAs by 7Æ5 ± 2Æ9& for diet and

7Æ3 ± 0Æ8& for fishmuscle.

Muscle essential AA d13C values showed stronger linear

relationships to their respective dietary AA d13C values with

slopes closer to unity (m = 0Æ9 ± 0Æ2, b = )2Æ7 ± 3Æ8,

Fig. 1. Mean (±SD) bulk tissue and individual amino acid d13C values of (a) diet and (b) Fundulus heteroclitus muscle from four dietarytreatments: Vegi-Pro (open squares), Bio-Vita (light gray triangles), squid (dark gray circles) and clam (black diamonds) (n = 5 replicates pertreatment for diets and n = 3 tanks per treatment, consisting of three fish per tank for fishmuscle).

Table 1. Proximate analysis of moisture, crude protein, crude fat,

crude fiber, ash and carbohydrate content (%) of four diets Vegi-Pro,

Bio-Vita, squid and clam (n = 1)

Analysis Vegi-Pro Bio-Vita Squid Clam

Moisture 6Æ8 6Æ0 10Æ0 8Æ8Protein (crude) 8Æ0 53Æ3 69Æ1 71Æ0Fat (crude) 5Æ9 23Æ9 17Æ6 18Æ0Fibre (crude) 2Æ0 0Æ3 0Æ3 0Æ2Ash 6Æ7 10Æ9 2Æ9 2Æ1Carbohydrates 72Æ6 6Æ0 0Æ3 0Æ2

Carbon isotope fractionation of fish muscle amino acids 1135

! 2010 TheAuthors. Journal compilation! 2010 British Ecological Society, Journal of Animal Ecology, 79, 1132–1141

r2 = 0Æ95 ± 0Æ04) than was found for non-essential AAs

(m = 0Æ51 ± 0Æ14, b = )7Æ0 ± 3Æ5, r2 = 0Æ71 ± 0Æ21)(Table 2). Muscle essential AA d13C values were also more

closely related to their dietary AA d13C values than they were

to bulk diet d13C values (m = 0Æ7 ± 0Æ4, b = )5Æ7 ± 11Æ7,r2 = 0Æ78 ± 0Æ18) (Table 2). Non-essential AAs d13C values

in muscle tissue typically showed stronger correlations to

bulk diet d13C values (m = 0Æ9 ± 0Æ5, b = 5Æ7 ± 15Æ3,r2 = 0Æ69 ± 0Æ26) than to their respective dietary AA d13Cvalues (Table 2).

Bulk fish muscle showed positive, albeit diet-specific, tro-

phic fractionation (Fig. 2) for all treatments, with Vegi-Pro

having the highest D13CC-D values and the squid and clam

treatments having the lowest D13CC-D values. There was a

large range in D13CC-D (12Æ5&) values across individual

AAs and dietary treatments (Fig. 2), from )7Æ9 ± 0Æ7& for

glycine in the Bio-Vita treatment to 4Æ7 ± 0Æ5& for glu-

tamic acid in the Vegi-Pro treatment. Even within individual

AAs, the mean range in D13CC-D across all four diets was

large (5Æ8 ± 2Æ9&), ranging from aspartic acid (3Æ5&) to

glycine (11Æ2&). While there was significant variability in

D13CC-D values among diets and individual AAs, we

observed several consistent patterns in the data. Essential

AA D13CC-D values were very consistent among individual

AAs (two-way anova, d.f. = 4, 40, F = 2Æ5, P > 0Æ05) anddietary treatments (d.f. = 3, 40, F = 1Æ7, P > 0Æ05). All

essential AA D13CC-D values were not significantly different

from 0& (mean D13CC-D = 0Æ02 ± 0Æ44&) (one sample t-

test, P > 0Æ05 for all essential AAs).

Conversely, D13CC-D values of non-essential AAs showed

much larger deviations from 0& and considerably more vari-

ation among AAs (two-way anova, d.f. = 5, 48, F = 165Æ0,

Table 2. Slope (m), y-intercept (b) and R2 values from linear regression analysis of Fundulus heteroclitus muscle amino acid d13C values

compared to (a) dietary amino acid d13C values and (b) bulk diet d13C values

Amino acid

(a) d13Cmuscle-AA vs. d13Cdiet-AA (b) d13Cmuscle-AA vs. d13Cbulk-diet

m b R2 m b R2

Glycinea 0Æ40 ± 0Æ01 )3Æ09 ± 0Æ23 0Æ86 ± 0Æ07 1Æ43 ± 0Æ09 26Æ33 ± 1Æ74 0Æ86 ± 0Æ07Serinea 0Æ62 ± 0Æ07 )2Æ92 ± 0Æ10 0Æ66 ± 0Æ01 1Æ73 ± 0Æ12 30Æ49 ± 2Æ26 0Æ89 ± 0Æ01Argininea 0Æ67 ± 0Æ21 )5Æ09 ± 3Æ22 0Æ41 ± 0Æ10 0Æ53 ± 0Æ16 1Æ05 ± 6Æ07 0Æ39 ± 0Æ09Aspartic acida 0Æ51 ± 0Æ11 )7Æ39 ± 1Æ67 0Æ77 ± 0Æ04 0Æ51 ± 0Æ12 )4Æ55 ± 2Æ54 0Æ55 ± 0Æ15Glutamic acida 0Æ45 ± 0Æ11 )8Æ66 ± 2Æ16 0Æ93 ± 0Æ05 0Æ75 ± 0Æ24 )1Æ72 ± 5Æ05 0Æ84 ± 0Æ21Prolinea 0Æ39 ± 0Æ10 )10Æ40 ± 1Æ81 0Æ46 ± 0Æ13 0Æ48 ± 0Æ13 )7Æ51 ± 2Æ80 0Æ35 ± 0Æ12Alaninea 0Æ54 ± 0Æ06 )11Æ03 ± 0Æ97 0Æ92 ± 0Æ03 0Æ73 ± 0Æ08 )4Æ34 ± 2Æ04 0Æ93 ± 0Æ04Threonineb 1Æ16 ± 0Æ10 1Æ82 ± 0Æ93 0Æ99 ± 0Æ01 1Æ15 ± 0Æ10 13Æ83 ± 1Æ94 0Æ83 ± 0Æ06Isoleucineb 0Æ77 ± 0Æ24 )4Æ92 ± 4Æ37 0Æ91 ± 0Æ02 0Æ46 ± 0Æ13 )9Æ98 ± 2Æ54 0Æ87 ± 0Æ11Valineb 0Æ87 ± 0Æ09 )2Æ84 ± 2Æ28 0Æ97 ± 0Æ05 1Æ10 ± 0Æ13 )1Æ14 ± 2Æ93 0Æ92 ± 0Æ11Phenylalanineb 0Æ88 ± 0Æ16 )3Æ19 ± 3Æ95 0Æ96 ± 0Æ04 0Æ60 ± 0Æ11 )13Æ47 ± 2Æ31 0Æ80 ± 0Æ10Leucineb 0Æ82 ± 0Æ17 )4Æ52 ± 4Æ35 0Æ95 ± 0Æ03 0Æ40 ± 0Æ10 )17Æ55 ± 1Æ99 0Æ47 ± 0Æ06MeanNEAAa 0Æ51 ± 0Æ14 )6Æ95 ± 3Æ49 0Æ71 ± 0Æ21 0Æ88 ± 0Æ49 5Æ68 ± 15Æ28 0Æ69 ± 0Æ26Mean EAAb 0Æ90 ± 0Æ20 )2Æ73 ± 3Æ84 0Æ95 ± 0Æ04 0Æ74 ± 0Æ35 )5Æ66 ± 11Æ71 0Æ78 ± 0Æ18

NEAAa = non-essential amino acids, and EAAb = essential amino acids (n = 3 tanks per treatment consisting of three fish per tank).

Fig. 2. Bulk tissue and individual amino acid stable carbon isotope trophic fractionation (D13CC-D ±SD) between diet and consumer forFundulus heteroclitus fed four dietary treatments: Vegi-Pro (open squares), Bio-Vita (light gray triangles), squid (dark gray circles) and clam(black diamonds) (n = 3 tanks per treatment, consisting of three fish per tank).

1136 K.W.McMahon et al.

! 2010 TheAuthors. Journal compilation! 2010 British Ecological Society, Journal of Animal Ecology, 79, 1132–1141

P < 0Æ05) and among treatments (d.f. = 3, 48, F = 218Æ4,P < 0Æ05), including a significant interaction term

(d.f. = 15, 48, F = 35Æ1, P < 0Æ05). Only arginine in the

Vegi-Pro treatment (one sample t-test, t = 0Æ8, P > 0Æ05)and glutamic acid in the Bio-Vita (one sample t-test, t = 0Æ3,P > 0Æ05), squid (t = 0Æ6, P > 0Æ05) and clam (t = 0Æ8,P > 0Æ05) treatments had D13CC-D values that were not

significantly different from 0&. Non-essential AA D13CC-D

values generally followed the patterns observed in the bulk

tissues D13C values (Fig. 2). Non-essential AA D13CC-D

values were typically the most positive in the Vegi-Pro treat-

ment with the exceptions of serine and arginine, where

Bio-Vita showed the highest D13CC-D values. Conversely,

D13CC-D values of non-essential AAs from the squid and clam

treatments were never significantly different from one

another (Tukey’s HSD post-hoc test, P > 0Æ05) and were

generally the lowest values.

There was notable variation in the per cent abundance of

AAs for both diets (Fig. 3a) and fishmuscle (Fig. 3b). In gen-

eral, the non-essential AAs glutamic acid, aspartic acid and

arginine were the most abundant AAs. Although lysine was

also quite abundant in both diet andmuscle, it was not analy-

sed for d13C due to coelution with tyrosine. Leucine was the

most abundant essential AA that was analysed for d13C. Thepatterns of per cent abundance of AAs were very consistent

across treatments for muscle (Fig. 2b), with a mean standard

deviation of 0Æ1 ± 0Æ1% across all AAs. There was consider-

ably more variation in AA per cent abundance across the

four diets (Fig. 2a), with a mean standard deviation of

1Æ0 ± 0Æ8% across all AAs. In the Vegi-Pro and Bio-Vita

treatments, all of the non-essential AAs analysed were less

abundant in the diets than they were in the muscle (mean dif-

ference in per cent abundance, Vegi-Pro: )2Æ0 ± 1Æ6% and

Bio-Vita: )1Æ5 ± 1Æ0%). The squid and clam diets usually,

but not exclusively, had a surplus of non-essential AAs

(0Æ0 ± 0Æ9% and 0Æ4 ± 1Æ2% respectively).

There was a significant negative correlation between the

difference in non-essential AA per cent abundance in diet

and muscle vs. AA D13CC-D (correlation coefficient,

r = )0Æ43, P < 0Æ05; Fig. 4). Biosynthesized non-essential

AAs tended to exhibit larger D13CC-D values when there was

a greater deficit in AA per cent abundance in the diet relative

Fig. 3. Mean per cent abundance (% ±SD) of 16 individual amino acids (left axis) and the total per cent abundance of the 12 amino acidsanalysed for d13C values (right axis) in (a) diet and (b) Fundulus heteroclitusmuscle from four dietary treatments: Vegi-Pro (open bars), Bio-Vita(light gray bars), squid (dark gray bars) and clam (black bars) (n = 3 replicates per treatment).

Carbon isotope fractionation of fish muscle amino acids 1137

! 2010 TheAuthors. Journal compilation! 2010 British Ecological Society, Journal of Animal Ecology, 79, 1132–1141

to fish muscle. The Bio-Vita treatment showed the most vari-

ability in D13CC-D values, with the D13CC-D values of aspartic

acid and proline in the Bio-Vita treatment closer those in the

Vegi-Pro treatment and the D13CC-D value of glycine more

similar to those in the natural animal diet treatments (Fig. 2).

Both aspartic acid and proline showed large deficits in the

Bio-Vita diet compared to fish muscle, as was the case for

Vegi-Pro (Fig. 3). Conversely, glycine was much closer to the

per cent abundance in muscle for the Bio-Vita, squid and

clam treatments (Fig. 3).

Discussion

We examined variability in carbon isotope fractionation of

individual AAs in a common marine fish across a wide

range of potential diets. Modest diet-specific D13CC-D values

in bulk tissue reflected relatively large trophic fractionations

for many non-essential AAs and little to no fractionation

for all essential AAs. Essential AA d13C values reflected a

purely dietary signature with D13CC-D values near 0&,

making them ideal tracers of carbon sources at the base of

the food web. Consumer non-essential AAs showed a large

range in D13CC-D across diets and a variable, diet-specific

degree of isotopic routing from dietary protein, which

together may contribute significantly to the high variability

in bulk tissue D13CC-D observed in the natural environment.

The diet-specific fractionation we found should promote

discussion about the assumptions of minimal and invariant

bulk tissue carbon isotope fractionation in dietary recon-

structions.

The patterns observed in the bulk d13C values (Elsdon

et al. 2010) were reflected in the d13C values of individual

AAs (Fig. 1). For instance, AAs from the Vegi-Pro treatment

(diet and muscle) were always the most 13C-depleted, while

those from the clam and squid treatments were typically the

most 13C-enriched. This is not surprising given that protein

was a significant component (up to 71%) of the diets

(Table 1) and fish muscle, making AAs a major contributor

to bulk tissue d13C values. There were several consistent pat-

terns in AA d13C values across all treatments. The large range

in AA d13C values of diet (27Æ9 ± 6Æ9&; Fig. 1a) and

consumer muscle (23Æ6 ± 2Æ6&; Fig. 1b) likely reflected the

varied metabolic histories of these AAs. These ranges were

similar to previous results on a variety of taxa including

vertebrates (Hare et al. 1991; Fogel & Tuross 2003; Howland

et al. 2003; Jim et al. 2006), invertebrates (Uhle et al. 1997;

Fantle et al. 1999; O’Brien, Fogel & Boggs 2002) and plants

(Fogel & Tuross 1999) from terrestrial and aquatic systems.

This consistency likely reflects the influence of two main fac-

tors: (i) similarities in the major biosynthetic pathways that

produce AAs in plants and animals and (ii) incorporation of

dietary constituents directly into consumer tissue.

The patterns of D13CC-D of individual AAs generally mir-

rored those of the bulk tissue (Fig. 2), with D13CC-D values in

the Vegi-Pro and Bio-Vita treatments significantly higher

than those in the squid and clam treatments. A closer look at

the D13CC-D values of individual AAs revealed some interest-

ing insights into metabolic processes impacting the synthesis

of muscle from dietary constituents. All essential AAs had

D13CC-D values near 0& (Fig. 2), indicating no significant

carbon isotope fractionation between diet and consumer

muscle AAs. This observation was supported by the

strong correlation and nearly 1 : 1 relationship between

d13Cdiet_essential_AA and d13Cmuscle_essential_AA (Table 2). Small

deviations from D13CC-D = 0&, and thus a slope of 1, most

likely represented minor kinetic isotope fractionation during

catabolism or conversion of essential AAs to other metabo-

lites. If we interpret the slope of this regression to be roughly

equivalent to the proportion of carbon routed into muscle

directly from the diet, the results support our hypothesis of a

high degree of isotopic routing of essential AAs into con-

sumer muscle. Our data support previous work on a variety

of taxa and tissues (Hare et al. 1991; Fantle et al. 1999; How-

land et al. 2003; O’Brien, Boggs & Fogel 2003; Jim et al.

2006), indicating that these findings are generally applicable

to a wide range of taxa and tissue types.

Although plants and bacteria can synthesize essential AAs

de novo, most animals have lost the necessary enzymatic

pathways to synthesize these AAs at a rate sufficient for nor-

mal growth, and thus must incorporate them directly from

their diet (Borman et al. 1946; Reeds 2000). As a result, the

d13C value of consumer essential AAs, such as phenylalanine

and leucine, must represent the isotopic fingerprint of pri-

mary producers at the base of the food web (d13CBase). It

should be noted that this relationship could be obscured

when dealing with strict herbivores that receive a significant

contribution of bacterially synthesized AAs from their sym-

biotic gut microbial community (Rimmer & Wiebe 1987).

The fidelity with which essential AAs reflect dietary sources

makes compound-specific SIA a powerful tool for foraging

ecology and dietary reconstruction. Essential AA d13C values

have provided insights into the diet of ancient humans and

Fig. 4. Differences between amino acid per cent abundance in dietand muscle (mean % ±SD) vs. stable carbon isotope trophicfractionation (D13CC-D ±SD). Negative values signify a lower percent abundance or d13C value in the diet relative to themuscle respec-tively (n = 3 for per cent abundance and n = 3 tanks per treatment,consisting of three fish per tank forD13CC-D).

1138 K.W.McMahon et al.

! 2010 TheAuthors. Journal compilation! 2010 British Ecological Society, Journal of Animal Ecology, 79, 1132–1141

herbivores (Stott et al. 1999; Fogel & Tuross 2003), the allo-

cation of adult resources to eggs in butterflies (O’Brien, Fogel

& Boggs 2002; O’Brien, Boggs & Fogel 2003, 2005), the con-

tributions of carbon sources tomarine dissolved organicmat-

ter (McCarthy et al. 2004), and the importance of marsh-

derived diets in supporting the growth of juvenile blue crabs

(Fantle et al. 1999). This approach may also provide a pow-

erful new tool for reconstructing the diet of highly mobile

consumers that move among isotopically distinct food webs.

Certainly compound-specific SIA avoids the confounding

variable of determining whether consumers with different

bulk tissue d13C values represent feeding in the same food

web but at different trophic levels, or feeding at the same

trophic level but in isotopically distinct food webs (Post 2002).

Non-essential AAs showed significant deviations from

D13CC-D = 0&, and much greater variability both among

AAs and across diet treatments compared to essential AAs

(Fig. 2). This variability most likely reflects the influence of

the varied metabolic processes that shape the isotopic signa-

tures of non-essential AAs during biosynthesis. The Vegi-Pro

treatment exhibited primarily positive D13CC-D values while

the natural diet treatments, squid and clam, typically showed

large negative D13CC-D values (Fig. 2). The large D13CC-D

values that shifted muscle non-essential AA d13C towards

bulk diet d13C values suggest a high degree of de novo biosyn-

thesis. This hypothesis was supported by linear regressions

between d13Cdiet_non-essential_AA and d13Cmuscle_non-essential_AA,

where the mean slopes were far from unity (Table 2), indicat-

ing a disparity between the d13C values of dietary and muscle

non-essential AAs.

The high degree of biosynthesis is surprising for the three

diets containing animal matter, Bio-Vita, squid and clam,

given the high protein content of those diets (53%, 69% and

71% respectively; Table 1). Previous research suggested that

when fed high protein diets, organisms typically route most

AAs, including non-essentials, directly from diet as a means

of energy conservation, because dietary routing is typically

more efficient than de novo biosynthesis (Ambrose & Norr

1993; Tieszen & Fagre 1993; Jim et al. 2006). Fish, however,

use a significant portion of dietary protein for energetic pur-

poses (Kim, Kayes & Amundson 1991; Dosdat et al. 1996),

and thus it is possible that fish exhibit a lower degree of die-

tary routing than terrestrial vertebrates. Only the Vegi-Pro

diet had a low protein content (8%) that would likely require

biosynthesis, resulting in the high D13CC-D observed across

most individual AAs in that treatment (Fig. 2). Hare et al.

(1991) found that d13C of proline and glutamate differed by

5Æ7& in the bone collagen of pigs, suggesting that proline was

being directly routed from diet into the consumer tissue. We

found that proline had d13C signatures closer to those of

glutamic acid rather than dietary proline (Fig. 1b) and had

D13CC-D values significantly different from 0& (Fig. 2). This

suggests that proline was biosynthesized from glutamic acid

via reduction through a Schiff base intermediate (Baich &

Pierson 1965) rather than being directly routed from the diet.

Arginine in the Vegi-Pro treatment and glutamic acid in

the Bio-Vita, squid and clam treatments showed strong evi-

dence of isotopic routing directly from dietary protein

(Fig. 2), yet evidence of biosynthesis in the other dietary

treatments. Arginine is synthesized from glutamate via glut-

amyl-c-semialdehyde and thus if arginine and glutamic acid

were both biosynthesized or both isotopically routed, we

would expect them to have similar d13C values, as was the

case for glutamic acid and proline discussed earlier. However,

arginine and glutamic acid had different d13C values, reflect-

ing the different pathways leading to arginine and glutamic

acid incorporation into fish muscle. Glutamic acid and argi-

nine account for over 18% of the AAs in fish muscle alone

(Fig. 3), and it is probable that otherAAs can be directly rou-

ted as well. We found that when glutamic acid was biosynthe-

sized in the Vegi-Pro treatment, it exhibited relatively large

D13CC-D values (!5&) similar to the D13CC-D values Hare

et al. (1991) and Howland et al. (2003) found for biosynthe-

sized glutamic acid in pig bone collagen (!6–7&). Thus vary-

ing degrees of isotopic routing vs. de novo biosynthesis for

these abundant AAs could significantly alter consumer tissue

d13C values relative to diet, further complicating the stable

isotope relationship between diet and consumer.

We hypothesized that an underrepresentation of non-

essential AAs in diet relative to muscle composition (Fig. 3)

would necessitate a higher degree of biosynthesis than would

be expected from diets with excess non-essential AAs. We

found a significant correlation between diet and muscle AA

per cent abundance and D13CC-D for biosynthesized non-

essential AAs (Fig. 4). The AA composition of fish muscle is

highly conserved (Wilson & Cowey 1985) as evidenced by the

fact that muscle AA per cent abundance was very consistent

across treatments (mean SD across treatments 0Æ1 ± 0Æ1%;

Fig. 3b) despite feeding on diets with highly variable AA con-

tent (1Æ0 ± 0Æ8%; Fig. 3a). When there was a deficit in AA

per cent abundance in the diet relative to the muscle, there

tended to be greater trophic fractionation. This was particu-

larly true for the Vegi-Pro and Bio-Vita treatments, where all

non-essential AAs analysed were less abundant in the diets

(Fig. 3a) than they were in fishmuscle (Fig. 3b), and typically

had the highest D13CC-D values. The disparity in AA per cent

abundance was perhaps not surprising given that Vegi-Pro

and Bio-Vita both contained plant matter, while the other

diets were entirely animal protein.

Bio-Vita showed the most variability in D13CC-D values

(Fig. 2), with some AAs trending towards Vegi-Pro and

other towards the squid and clam treatments. For example,

aspartic acid and proline had similar D13CC-D values in the

Bio-Vita and Vegi-Pro treatments. Those AAs also showed

large deficits in the Bio-Vita and Vegi-Pro diets compared to

fish muscle. Conversely, glycine in the Bio-Vita treatment

had a very negativeD13CC-D value closer to those of the squid

and clam treatments. In this case, the disparity in glycine per

cent abundance between diet and fish muscle was small for

the Bio-Vita, squid, and clam treatments. Differences in AA

abundance in the diet relative to consumer muscle likely

required varying the degree of biosynthesis and catabolism to

meet the muscle composition demand, which may explain the

corresponding shifts in AA trophic fractionation. However,

Carbon isotope fractionation of fish muscle amino acids 1139

! 2010 TheAuthors. Journal compilation! 2010 British Ecological Society, Journal of Animal Ecology, 79, 1132–1141

disparities in diet and muscle AA composition alone only

explain a relatively small fraction (r2 = 0Æ19) of D13CC-D val-

ues.

The differences in D13CC-D values between the Vegi-Pro

treatment and the squid and clam treatments may reflect dif-

ferences in utilization of the bulk carbon pool from a plant-

based diet vs. an animal-based diet. The Vegi-Pro diet had far

more carbohydrates (73%) than lipids (6%), while the ani-

mal-based diets showed the opposite trend (18% lipid, <1%

carbohydrate (Table 1). The biosynthesis of non-essential

AAs in the Vegi-Pro treatment appeared to rely on a more13C-enriched carbon pool than the other treatments, possibly

indicating a greater contribution of carbohydrates to the

bulk carbon pool (Teece & Fogel 2007). Howland et al.

(2003) reared pigs on a plant-based diet with a d13C value

close to the Vegi-Pro diet used in our study. Our results were

similar to those of pig collagenD13CC-D values, showing large

positive D13CC-D values for most non-essential AAs, particu-

larly glutamic acid, proline and aspartic acid. Similar

metabolic processes may, therefore, control D13CC-D values

for many animals feeding on plant-based diets.

If lipids in the animal-based diets were being catabolized as

a significant source of energy (Post et al. 2007), they would

provide a very 13C-depleted carbon pool relative to bulk diet

values (6–8&; DeNiro & Epstein 1977) from which non-

essential AAs were biosynthesized. This may explain why the

D13CC-D values in the animal-based dietary treatments were

significantly more negative than in the Vegi-Pro treatment

(Fig. 2). The divergence in D13CC-D between Vegi-Pro and

the squid and clam treatments is greatest for glycine, serine

and alanine, which are also the first AAs synthesized from

carbohydrates entering the glycolysis as glucose. Glucose is

converted to 3-phosphogylcerate, which is the precursor for

both glycine and serine. Alanine is synthesized from pyruvate

several steps after 3-phosphogylcerate and showed less of a

difference in D13CC-D between the plant- and animal-based

diets. The remaining non-essential AAs are synthesized from

oxaloacetate and a-ketogluterate intermediates many steps

later in the TCA cycle and showed the smallest differences in

D13CC-D between the plant- and animal-based diets. If differ-

ent carbon pools are in fact driving the diet-specific differ-

ences in D13CC-D values of non-essential AAs, the impact

appears to be greatest near the source of carbon entering gly-

colysis and gets diluted or altered as carbon flows through

the TCA cycle. Our work supports previous observations

that organisms feeding on apparently homogeneous diets can

show substantially different d13C values when routing of

dietary components and alterations of available carbon pool

d13C values become important (O’Brien, Fogel & Boggs

2002; O’Brien, Boggs &Fogel 2003; Jim et al. 2006).

The diets chosen for this study ranged from solely plant

matter to solely animal matter to examine the potential vari-

ability in diet to consumer carbon isotope fraction. Without

knowing the fractionation between steps as lipid and carbo-

hydrate carbon enter the TCA cycle and get incorporated

into AAs, we cannot accurately predict how the precursor

d13C signatures will be manifested in the product AA d13C

values. The next step will be to identify the mechanisms

behind the high, diet-specific variability in D13CC-D and

determine what information non-essential AA d13C values

hold about consumer diet and metabolic history. This calls

for targeted feeding experiments that track the fractionation

of individual, potentially isotopically labelled dietary constit-

uents as they are metabolically processed. While it is cur-

rently unclear how much useful information about diet and

metabolic history is recorded in non-essential AA d13C val-

ues, the fact that theD13CC-D values in both animal diet treat-

ments tracked very closely and were always significantly

different from the plant-based Vegi-Pro diet (Fig. 2) holds

promise that there may be some valuable underlying princi-

ples controlling consumer individual AA d13C values.

Acknowledgements

The authors would like to thank S. Ayvazian, D. Champlin, J. Serbert and B.Miner of the Atlantic Ecology Division, U.S. Environmental ProtectionAgency, in Narragansett, Rhode Island for assistance rearing fish and L.Houghton for assistance in sample analysis. The authors would also like tothank the Mullineaux lab group for comments on an early draft of the manu-script andM.McCarthy for a thorough review of this manuscript. The feedingexperiment was supported by the Atlantic Ecology Division, U.S. Environ-mental Protection Agency, and funding from the Ocean Life Institute and anAustralian Research Council fellowship to T. Elsdon. Additional funding wasprovided by the University of New Hampshire’s Large Pelagics Research Cen-ter, Woods Hole Oceanographic Institution, the Carnegie Institution ofWash-ington, and theW.M. Keck Foundation. K.McMahon received support fromtheNational Science FoundationGraduate Research Fellowship Program andT. Elsdon received support from the Post Doctoral Scholar Program atWoodsHole Oceanographic Institution.

References

Ambrose, S.H. & Norr, L. (1993) Carbon isotopic evidence for routing ofdietary protein to bone collagen, and whole diet to bone apatite carbonate:purified diet growth experiments. Molecular Archaeology of PrehistoricHuman Bone (eds J. Lambert & G. Grupe), pp. 1–37. Springer-Verlag,Berlin.

AOAC (2005) Official Methods of Analysis, 17th edn. AOAC International,Arlington, VA,USA.

Baich, A. & Pierson, D.J. (1965) Control of proline synthesis inEscherichia coli.Biochimica et Biophysica Acta, 104, 397–404.

Borman, A., Wood, T.R., Black, H.C., Anderson, E.G., Oesterling, M.J.,Womack, M. & Rose, W.C. (1946) The role of arginine in growth with someobservations on the effects of argininic acid. Journal of Biological Chemistry,166, 585–594.

Cherel, Y., Hobson, K.A., Guinet, C. & Vanpe, C. (2007) Stable isotopesdocument seasonal changes in trophic niches and winter foraging individualspecialization in diving predators from the Southern Ocean. Journal ofAnimal Ecology, 76, 826–836.

DeNiro, M.J. & Epstein, S. (1977) Mechanism of carbon isotope fractionationassociated with lipid synthesis.Science, 197, 261–263.

DeNiro, M.J. & Epstein, S. (1978) Influence of diet on distribution of carbonisotopes in animals.Geochimica et Cosmochimica Acta, 42, 495–506.

Dosdat, A., Servais, F., Metailler, R., Huelvan, C. & Desbruyeres, E. (1996)Comparison of nitrogenous losses in five teleost fish species. Aquaculture,141, 107–127.

Elsdon, T., Ayvazian, S., McMahon, K.M. & Thorrold, S.R. (2010) Experi-mental evaluation of stable isotope fractionation in fish muscle and otoliths.Marine Ecology Progress Series, 404, 195–205.

Estrada, J.A., Lutcavage,M.&Thorrold, S.R. (2005) Diet and trophic positionof Atlantic bluefin tuna (Thunnus thynnus) inferred from stable carbon andnitrogen isotope analysis.Marine Biology, 147, 37–45.

Fantle, M.S., Dittel, A.I., Schwalm, S.M., Epifanio, C.E. & Fogel, M.L. (1999)A food web analysis of the juvenile blue crab, Callinectes sapidus, usingstable isotopes in whole animals and individual amino acids.Oecologia, 120,416–426.

1140 K.W.McMahon et al.

! 2010 TheAuthors. Journal compilation! 2010 British Ecological Society, Journal of Animal Ecology, 79, 1132–1141

Fogel, M.L. & Tuross, N. (1999) Transformation of plant biochemicals togeological macromolecules during early diagenesis. Oecologia, 120, 336–346.

Fogel, M.L. & Tuross, N. (2003) Extending the limits of paleodietary studies ofhumans with compound specific carbon isotope analysis of amino acids.Journal of Archaeological Science, 30, 535–545.

Fry, B., Joern, A. & Parker, P.L. (1978) Grasshopper food web analysis: use ofcarbon isotope ratios to examine feeding relationships among terrestrialherbivores.Ecology, 59, 498–506.

Gannes, L.Z., Martınez del Rio, C. & Koch, P. (1998) Natural abundancevariations in stable isotopes and their potential uses in animal physiologicalecology.Comparative Biochemistry and Physiology, 119A, 725–737.

Gannes, L.Z., O’Brien, D.M. & Martınez del Rio, C. (1997) Stable isotopes inanimal ecology: assumptions, caveats, and a call for more laboratory experi-ments.Ecology, 78, 1271–1276.

Graham, B.S., Koch, P.L., Newsome, S.D., McMahon, K.W. & Aurioles, D.(2009) Using isoscapes to trace the movements and foraging behavior of toppredators in oceanic ecosystems. Isoscapes: Understanding Movement,Pattern and Process on Earth Through Isotope Mapping (eds J. West, G.J.Bowen, T.E. Dawson & K.P. Tu). pp. 299–318. Springer, New York, NY,USA.

Hansson, S., Hobbie, J.E., Elmgren, R., Larsson, U., Fry, B. & Johansson, S.(1997) The stable nitrogen isotope ratio as a marker of food web interactionsand fishmigration.Ecology, 78, 2249–2257.

Hare, P.E., Fogel, M.L., Stafford, T.W., Mitchell, A.D. & Hoering, T.C.(1991) The isotopic composition of carbon and nitrogen in individual aminoacids isolated from modern and fossil proteins. Journal of ArchaeologicalScience, 18, 277–292.

Hobson, K.A. (1999) Tracing origins and migration of wildlife using stableisotopes: a review.Oecologia, 120, 314–326.

Howland, M.R., Corr, L.T., Young, S.M., Jones, V., Jim, S., Van der Merwe,N.J., Mitchell, A.D. & Evershed, R.P. (2003) Expression of the dietaryisotope signal in the compound-specific d13C values of pig bone lipids andamino acids. International Journal of Osteoarchaeology, 13, 54–65.

Jim, S., Jones, V., Ambrose, S.H. & Evershed, R.P. (2006) Quantifying dietarymacronutrient sources of carbon for bone collagen biosynthesis usingnatural abundance stable carbon isotope analysis. British Journal ofNutrition, 95, 1055–1062.

Karasov, W.H. &Martınez del Rio, C. (2007)Physiological Ecology. PrincetonUniversity Press, Princeton,New Jersey, USA.

Kim, K., Kayes, T.B. &Amundson, C.H. (1991) Purified diet development andre-evaluation of the dietary protein requirement of fingerling rainbow trout(Onchorynchusmykiss).Aquaculture, 96, 57–67.

McCarthy, M.D., Benner, R., Lee, C., Hedges, J.I. & Fogel, M.L. (2004)Amino acid carbon isotope fractionation patterns in oceanic dissolvedmatter: an unaltered photoautotrophic source for dissolved organic nitrogenin the ocean?Marine Chemistry, 92, 123–134.

McCutchan, J.H. Jr, Lewis, W.M. Jr, Kendall, C. & McGrath, C.C. (2003)Variation in trophic shift for stable isotope ratios of carbon, nitrogen, andsulfur.Oikos, 102, 378–390.

McMahon, K.W., Johnson, B.J. & Ambrose, W.G. (2005) Diet and movementof the killifish, Fundulus heteroclitus, in a Maine salt marsh assessed usinggut contents and stable isotope analyses.Estuaries, 28, 966–973.

Meier-Augenstein, W. (1999) Applied gas chromatography coupled to isotoperatiomass spectrometry. Journal of Chromatography, 842, 351–371.

Merritt, D.A., Brand, W.A. & Hayes, J.M. (1994) Isotope-ratio-monitoringgas chromatography-mass spectrometry: methods for isotopic calibration.Organic Geochemistry, 21, 573–583.

Michener, R.H. & Kaufmann, L. (2007) Stable isotope ratios as tracers inmarine aquatic food webs: an update. Stable Isotopes in Ecology andEnvironmental Science, 2nd edn (eds K. Lajtha & R.H. Michener), pp. 238–282. Blackwell Publishing,Malden,MA,USA.

O’Brien, D.M., Boggs, C.L. &Fogel,M.L. (2003) Pollen feeding in the butterflyHeliconius charitonia: isotopic evidence for essential amino acid transferfrom pollen to eggs. Proceedings of the Royal Society of London. Series B,Biological Sciences, 270, 2631–2636.

O’Brien, D.M., Boggs, C.L. & Fogel, M.L. (2005) The amino acids used inreproduction by butterflies: a comparative study of dietary sources using

compound-specific stable isotope analysis. Physiological and BiochemicalZoology, 78, 819–827.

O’Brien, D.M., Fogel, M.L. & Boggs, C.L. (2002) Renewable and nonrenew-able resources: amino acid turnover and allocation to reproduction inLepidoptera. Proceedings of the National Academy of Sciences, USA, 99,4413–4418.

Olive, P.J.W., Pinnegar, J.K., Polunin, N.V.C., Richards, G. & Welch, R.(2003) Isotope trophic-step fractionation: a dynamic equilibrium model.Journal of Animal Ecology, 72, 608–617.

Parkington, J. (1991) Approaches to dietary reconstruction in the WesternCape: Are you what you have eaten? Journal of Archaeological Science, 18,331–342.

Peterson, B.J. & Fry, B. (1987) Stable isotopes in ecosystem studies. AnnualReview of Ecology and Systematics, 18, 293–320.

Popp, B.N., Graham, B.S., Olson, R.J., Hannides, C.C.S., Lott, M.J.,Lopez-Ibarra, G.A., Galvan-Magana, F. & Fry, B. (2007) Insight into thetrophic ecology of yellowfin tuna, Thunnus albacares, from compound-specific nitrogen isotope analysis of proteinaceous amino acids. StableIsotopes as Indicators of Ecological Change (eds T.D. Dawson & R.T.W.Siegwolf), pp. 173–190. Elsevier ⁄Academic Press, Amsterdam.

Post, D.M. (2002) Using stable isotopes to estimate trophic position: models,methods, and assumptions.Ecology, 83, 703–718.

Post, D.M., Layman, C.A., Arrington, D.A., Takimoto, G., Quattrochi, J. &Montana, C.G. (2007) Getting to the fat of the matter: models, methods andassumptions for dealing with lipids in stable isotope analysis.Oecologia, 152,179–189.

Reeds, P. (2000) Dispensable and indispensable amino acids for humans.Journal of Nutrition, 130, 1835S–1840S.

Rimmer, D.W. & Wiebe, W.J. (1987) Fermentative microbial digestion inherbivorous fish. Journal of Fish Biology, 31, 229–236.

Rubenstein, D.R. & Hobson, K.A. (2004) From birds to butterflies: animalmovement patterns and stable isotopes. Trends in Ecology and Evolution, 19,256–263.

Schwarcz, H.P. (1991) Some theoretical aspects of isotope paleodiet studies.Journal of Archaeological Science, 18, 261–275.

Schwarcz, H.P. & Schoeninger, M.J. (1991) Stable isotope analyses in humannutritional ecology.Yearbook Physical Anthropology, 34, 283–321.

Sessions, A.L. (2006) Isotope-ratio detection for gas chromatography. Journalof Separation Science, 29, 1946–1961.

Silfer, J.A., Engel, M.H., Macko, S.A. & Jumeau, E.J. (1991) Stable carbonisotope analysis of amino-acid enantiomers by conventional isotope ratiomass spectrometry and combined gas-chromatography isotope ratiomass-spectrometry.Analytical Chemistry, 63, 370–374.

Stott, A.W., Evershed, R.P., Jim, S., Jones, V., Rogers, J.M., Tuross, N. &Ambrose, S.H. (1999) Cholesterol as a new source of palaeodietaryinformation: experimental approaches and archaeological applications.Journal of Archaeological Science, 26, 705–716.

Teece, M.A. & Fogel, M.L. (2007) Stable carbon isotope biogeochemistryof monosaccharides in aquatic organisms and terrestrial plants. OrganicGeochemistry, 38, 458–473.

Tieszen, L.L. & Fagre, T. (1993) Effects of diet quality and composition onisotopic composition of respiratory CO2, bone collagen, bioapatite, and softtissues. Prehistoric Human Bone Archaeology at the Molecular Level (edsJ. Lambert &G.Grupe), pp. 121–156. Springer-Verlag, Berlin.

Uhle, M.E., Macko, S.A., Spero, S.A., Engel, M.H. & Lea, D.W. (1997)Sources of carbon and nitrogen in modern plankton foraminifera: the role ofalgal symbionts as determined by bulk and compound specific stable isotopeanalysis.Organic Geochemistry, 27, 103–113.

Vander Zanden, M.J. & Rasmussen, J.B. (1999) Primary consumer d13C andd15N and the trophic position of aquatic consumers.Ecology, 80, 1395–1404.

Vander Zanden, M.J. & Rasmussen, J.B. (2001) Trophic fractionation:implications for aquatic food web studies. Limnology and Oceanography, 46,2061–2066.

Wilson, R.P. & Cowey, C.B. (1985) Amino acid composition of whole bodytissue of rainbow trout andAtlantic salmon.Aquaculture, 48, 373–376.

Received 1 September 2009; accepted 9 June 2010Handling Editor: RyanNorris

Carbon isotope fractionation of fish muscle amino acids 1141

! 2010 TheAuthors. Journal compilation! 2010 British Ecological Society, Journal of Animal Ecology, 79, 1132–1141